Thanks to everyone for their suggestions.
I think I'll try the "short rest = 8 hours and Long Rest = Suitable Downtime" approach.
I've read in multiple different places that one of the core assumptions in D&D 5e is a workday of around 6 encounters (combat or some other obstacles that consume resources) per day.
That's the best work-around I've seen: Don't just turn on one alternate rest variant, that only shifts the problem. Change the needed conditions/durations of rests to suit the current story dynamics, so the pacing always approximates the 6-8 encounter/2-3 short rest balance point at which classes are theoretically competitive with eachother, and encounters perform as close as possible to their ratings.The true solution is to have rest frequencies vary depending on the needs of the scenario.
This is the solution. Anything short of this does not solve the problem for all kinds of adventures.
Unfortunately, that's a commonly spread misconception.
The DMG suggest 6-8 encounters with 2 short rests, about 1/3 and 2/3 of the way through the day. (DMG pg 84.)
DMG said:Assuming typical adventuring conditions and average luck, most adventuring parties can handle about six to eight medium or hard encounters in a day. If the adventure has more easy encounters, the adventurers can get through more. If it has more deadly encounters, they can handle fewer.
Just having harder encounters doesn't balance because of durations. An easy example is the barbarian - starting at 2 rages a day, they are supposed to be able to rage 1/4 to 1/3 of the encounters. With 1-2 encounters per day they would always be able to rage - you'd need to scale it back to one rage every two or three days to hit the same balance point.
Look at spells. If encounters are fewer, any "all encounter" buffs will be great, and you're casting then 1-2 instead of 6-8 times. If encounters are harder because of mroe foes, area effect can often hit more and be a lot more effective. If combats are harder because of more dangerous opponents, then debuffs/crowd control can make a huge impact - and because of one 1/3 of the saves are proficient it will still be easy to find spells that will affect them even if they are tougher.
Fewer, harder encounters changes the balance dramatically vs. the expected amount. It's far from an unplayable balance, but it makes some classes a lot better, especially full casters, and makes other classes relatively weaker, like those that are primarily at-will (weapons, etc.)
A better solution is to scale back on your rests so that you can have the recommended number of encounters between them. The DMG has a Gritty Realism variant (pg 267) that makes a short rest 8 hours, and a long rest 7 days. So you can have 2 encounters per day and a short rest, and long rests only between adventures.
I have an idea about slower spell recovery as well, maybe you can recover a percentage (33%?) of your total spell levels per day. A 5th level Wizard with 5/3/2 spells has a total of 17 spell levels and would recover 6 levels worth of spells after a long rest.
Actually, the thing nobody seems to want to admit is:[MENTION=20564]Blue[/MENTION] I generally find your posts insightful and interesting; I hope you will not be offended by the directness of my comments below.
Actually, that is the commonly spread misconception. Here is the actual quote from the DMG.
So, first, the 6-8 number is associated with encounters of a particular difficulty, and the text explicitly mentions more and fewer encounters as equally viable alternatives. Per this section in the DMG, there is nothing special about 6-8 encounters. Second, this section in the DMG is not, per se, making a recommendation about how many encounters a party ought to have; rather, it is suggesting an approximate upper limit on what they can handle.
These are good points about how the number of encounters per "adventuring day" affects class balance. Except that there is no evidence to suggest that, for instance, a 1st level barbarian being able to rage in 1/4 to 1/3 of encounters is the intended or preferred balance point.
I have considered using this variant, but the notion that you can sit around for 6 days and not recover any resources and then suddenly be made whole after day 7 is just a little too weird (for me anyway). Admittedly, it is logically no different than the miraculous 8th hour of a standard long rest, but stretching out the fictional time just makes it seem way more discomfiting to me.
What I think I really would like is a finer-grained recovery over the extended time, but I haven't thought of a way to do that that isn't way fiddly.
Thanks to everyone for their suggestions.
I think I'll try the "short rest = 8 hours and Long Rest = Suitable Downtime" approach.